Speaker: Alexander Hartemink
Time and Location: 4:15 PM, B11 Kimball Hall
Title: Principled Computational Methods for the Validation and Discovery of Genetic Regulatory
Networks from Expression Data
With the Human Genome Project drawing near completion, the grand challenge in biology over the next twenty years will be to determine the roles of the thousands of genes inside every living cell. Interestingly (and fortunately for computer scientists), making progress in functional genomics is at least as much of a computational challenge as it is a biological one. My goal in this talk will be twofold: first, to convince you of both the importance and fertility of this area for computational research, and second, to offer some of my results indicating the kinds of contributions that computer scientists can make to this nascent field.
In this particular talk, I will first give an overview of functional genomics and then present a computational framework for the validation and discovery of genetic regulatory networks from expression data. This framework is based on existing work in graphical models, but includes extensions I have developed for modeling in the specific context of transcriptional regulation. I will discuss why graphical models are a suitable language for representing genetic regulatory networks and will present a series of results on appropriate normalization of expression array data, a method for information-preserving quantization of normalized data, and principled scoring of models of regulatory networks. I will also demonstrate how these methods can be used to score various hypotheses regarding the structure and function of genetic regulatory networks in the presence of expression data.
This talk is designed to be accessible to everyone in the computer science
community: no prior biological knowledge will be assumed or required.